Comprehensive Analysis of In-Car Voice Recognition Testing

Comprehensive Analysis of In-Car Voice Recognition Testing
From bench to actual vehicle, this voice recognition special testing practical notes are very detailed. Follow along to learn and understand!
1. Principles of Voice Recognition and Testing Scope
1. Principles of Voice Recognition:
① Input human voice through the microphone.
② Acoustic processing: eliminate noise and disturbances.
③ Feature processing: extract key factors from the voice, such as: 小米 (Xiaomi).
④ Model matching: for example, ‘xiaomi’ can match ‘小米’ or ‘小蜜’, and the system needs to calculate the most probable output based on the surrounding context for user confirmation.
2. Analysis of Testing Content, divided into two main phases: audio capture and processing feedback based on the above processing steps.
① Microphone input: such as Mandarin, speech rate, dialect, volume, timbre (male/female), network, noise, disturbance, interface voice icon display changes.
② Functionality and business processing: after capturing audio, convert it to text for display and perform corresponding operations based on the content.
A. Control the vehicle–testing in actual vehicle environment.
B. Control the APP (navigation, multimedia, settings, etc.).
C. Free dialogue (e.g., search knowledge, check weather, play games).
2. Bench Voice Testing Analysis
1. Bench Environment
Involves instruments, power supply, multimeters, oscilloscopes, etc., built and maintained according to the testing business and objects.
2. Control APP Analysis
Core: Recognizing keywords.
(1) Navigation Function
* I want to go to…
* Go to…
* …how to get there
* Navigate to…
* I want to go to….
(2) Phone

* Contact xxx

* Dial xxx phone

* Dial xxxx mobile/telecom/unicom number

* Dial home/company/family

* Dial the last three digits xxx/number starting with 187

* Dial Su San in Beijing/Shanghai/Hefei

* Call back xxx

* Contact/dial 187xxxx (specific number)

* Dial SF/mobile/KFC number (official number)….

(3) Multimedia

* Play music

* Play video

* Adjust volume, brightness

* Playback status, song switching…

(4) Weather, Knowledge Q&A

* Tomorrow’s weather (date, day)

* Nanjing weather (region)

* Knowledge Q&A…

(5) Combined Testing

* Play music, dial phone

* Answer phone during navigation

* Play music/video while navigating……

3. Interpretation of In-Car Voice Functions and Testing Explanation

In the actual vehicle testing part, besides basic verification of bench functions, the main focus is verifying vehicle control:

* Sunroof, windows, reading lights, tailgate (trunk), seats (ventilation, heating, backrest)……

* Combining gestures and voice (pointing to the window and saying: close/open it)……

Note: Specific business involves referring to the vehicle product user manual or the introduction of voice-related settings in the settings menu.

4. Organizing Voice Special Testing Content and Key Points

1. Key Testing Points for Voice Recognition in Bench and Actual Vehicle:

(1) Verification of bench environment—the number of use cases is comprehensive, mainly focusing on validating the voice recognition ‘keywords’.

(2) Actual vehicle—the central control voice control accounts for a smaller proportion, requiring basic function verification, considering both stationary and driving conditions, with the remainder focusing on vehicle control testing.

2. For voice recognition testing, consider the following:

(1) Recognition rate: whether different timbres and contents can be recognized and processed.

(2) Accuracy: the recognition accuracy of multiple, continuous, different keywords and descriptions, providing correct text feedback for user confirmation and corresponding correct action operations.

(3) Response time: verifying recognition and response time in different scenarios.

(4) Testing while driving: (multimedia, navigation, settings, environmental changes).

(5) Voice recognition functionality from different positions inside the vehicle.

5. Conclusion

In conclusion, to effectively test the business and functionality of in-car voice recognition, one must understand the principles of voice recognition, the recognition and processing flow, as well as the specific application methods and modules in the vehicle.

Consideration must also be given to the specific driving environment, such as changes in stationary and driving conditions. As the sound capture end, aspects such as sound source, timbre, angle, position, and tone must also be tested.

Comprehensive Analysis of In-Car Voice Recognition Testing
END

Leave a Comment